lmprices

AnthropicClaude Sonnet 3.7 API pricing

Claude Sonnet 3.7 was retired on . Its last price was $3.00 per 1M input tokens and $15.00 per 1M output tokens.

Last price

Rates in effect since (prices added). Every price links to its source.

Input
$3.00
Output
$15.00
Blended 3:1
$6.00
Claude Sonnet 3.7 last rates per 1M tokens
RatePer 1M tokens
Standard
Input$3.00
Cached input$0.30
Cache write (5 min)$3.75
Cache write (1 hour)$6.00
Output$15.00
Batch
Input$1.50
Output$7.50

Launch price

Claude Sonnet 3.7 launched at this price on .

Input
$3.00
Output
$15.00
Blended 3:1
$6.00

Price history

InputOutput
$0$5$10$15Mar 2025May 2025Jul 2025Sep 2025Nov 2025Jan 2026 Input $3.00, from February 24, 2025Output $15.00, from February 24, 2025Output $15.00Input $3.00 $0$5$10$15Jul 2025Jan 2026 Input $3.00, from February 24, 2025Output $15.00, from February 24, 2025Output $15.00Input $3.00
Headline standard prices per 1M tokens. A filled circle marks a dated change; an open circle, a change first seen between two captures.

Every price event

2 verified events, newest first. No price moves. Documented since .

  1. Prices added ·

    • Cache write (1 hour) added at $6.00
    Rates from May 22, 2025
    RatePer 1M tokens
    Standard
    Input$3.00
    Cached input$0.30
    Cache write (5 min)$3.75
    Cache write (1 hour)$6.00
    Output$15.00
    Batch
    Input$1.50
    Output$7.50

    1-hour cache writes added at $6 (2x base input); other rates unchanged. The May 22, 2025 announcement introduced the 1-hour cache TTL 'at an additional cost' without naming the price or the models; the price and the per-model scope come from the docs pricing table captured 2025-05-26. The docs pricing table captured 2025-05-13 has no 1-hour column. The 1-hour TTL was in beta until the beta header was dropped on 2025-08-13.

  2. Launch price ·

    Rates from February 24, 2025
    RatePer 1M tokens
    Standard
    Input$3.00
    Cached input$0.30
    Cache write (5 min)$3.75
    Output$15.00
    Batch
    Input$1.50
    Output$7.50

    Thinking tokens are billed as output at the same $15 rate, so thinking output isn't recorded separately. Batch rates come from the batch docs table captured 2025-02-25, the day after launch, which also lists Claude 3.7 Sonnet (claude-3-7-sonnet-20250219) as a supported model; the pricing page card the same day says '50% discount with batch processing'.

Gaps in this history

None known for input, output, cache and batch rates.

API ids

Checked .